def get_shop_top_terms(shop, etsy_api, term_limit=5, min_request_duration=0):
"""Extract the most popular terms for a single shop
shop (dict): etsy shop definition
etsy_api (EtsyAPI): Etsy API client instance
term_limit (int): number of top terms to display
"""
offset = 0
texts = []
while True:
start_time = time.time()
listings = etsy_api.get_shop_listings(shop["shop_id"],
limit=100,
offset=offset)
for listing in listings["results"]:
texts.extend([listing["title"], listing["description"]])
if not listings["pagination"]["next_page"]:
break
offset += 100
duration = time() - start_time
if duration < min_request_duration:
sleep(min_request_duration - duration)
terms = rank_terms(texts)
if terms is None:
return
top_terms = ''
for i in xrange(min(len(terms), term_limit)):
top_terms += str(i + 1) + "-" + terms[i][0] + " "
print "shop id %s: %s" % (shop["shop_id"], top_terms)
def test_rank_terms(self):
moby_dick = gutenberg.sents("melville-moby_dick.txt")
terms = rank_terms([" ".join(s) for s in moby_dick[:500]])
assert type(terms[0][0]) == unicode
assert type(terms[0][1]) == int
assert terms[0][0] == "whale"
